Likes and dislikes - Forum - OpenEdge Development - Progress Community
 Forum

Likes and dislikes

  • Why would OEA cripple basic functionality like

    that?

    We don't. The preference page action on the editor's

    context menu is put there by eclipse, not by us.

    Ok, I misread your original response.

    If this is Eclipse's doing, it's - well - lame and annoying. It should expand the Editor branch if it's not there already, put the cursor there, and highlight it. Not filter everything else out.

    We could replace/change/remove it,

    but there hasn't been any need.

    I think this qualifies as a need, particularly since the OE editor settings, keyboard key assignments, and such aren't considered "part" of the Eclipse editor configuration.

    We add contributions such as compile/check

    syntax and the source submenu, but the navigation and

    some of the other stuff is on there already.

    Ok.

    >> I've confirmed the flag is turned on. I'm still

    >> not getting auto-completion.

    Auto completion or context asist?

    auto completion.

    Auto completion

    can be turned on/off from

    window -> preferences -> general ->

    OEA -> editor -> assistance.

    Defaults to on.

    Correct.

    Context assist is ctrl+space.

    CA is ctl+space? Then what's auto-completion?

    The "assistance" preference says that CA is floating text while entry, while auto completion is ".", ":". (There's also "ctl+space will still work").

    auto completion

    shouldn't work on amibigous names.

    It shouldn't auto-complete when the current prefix is matched by multiple previous entries. It should provide a list of possible value to choose from though.

    Does your variable have the same starting

    character sequence as a language keyword

    or other field/variable?

    DEFINE QUERY q-name

    FOR rsinv.

    DEFINE BROWSE b-name

    QUERY q-name

    I was trying to get it to auto-complete the "q-name" on the query phrase of the browse statement. It wouldn't.

    >>So that what those two un-labelled numbers at the

    >>bottom are!

    yeah....again this is a case of eclipse provided

    functionality that we get for free.

    It needs an upgrade.

    >> For the editor templates, it looks like I can

    >> change the description and pattern, but I

    >> can't change the "name" - which I'm assuming

    >> is the keyboard sequence that fires each template

    >>off.

    Not sure it crossed anyone's mind to let it be

    modifiable.

    I find that rather astounding.

    No technical reason they can't be.

    There's good reason to make it so.

    >>The "edit template" window "pattern" editing

    >>area's awfully small. I've got some other templates

    >>that are a few K in size,

    A few k? wow.

    They're mostly skeletons for various ABL declarations that write the definition in a standardized format that I can then fill-in-the-blanks with.

    >>Ideally, templates should be edited in an editor

    >>window, unless they're supposed to be restricted

    >>to simplistic variable declarations.

    There is no arbtrary size restriction since its just

    a string in an xml file.

    note to self...make template macro window bigger. It

    is resizable (at least in 10.1B which is what I have

    in front of me).

    Excellent.

    Can templates be chained? Ie, one template invokes another one - maybe something like:

    &DVCH\n&DVI\n

    to declare a character var and then an integer var?

    >> What I'm thinking of is a file got into a project

    >> such as by the import process - that I decide I

    >>don't want in there - even if it is hidden behind

    >>a filter condition.

    >>

    >> I can't just dis-associate it from the project, I

    >> have to delete the file. While I can see the logic

    >> of associating those actions, I can think of cases

    >> where it's not desirable behavior.

    Import is just a copy operation as far as Eclipse is

    concerned.

    Importing a directory tree into a project is a copy operation - to where?

    Since it doesn't keep track of all the

    contents of a project in a separate file (unlike VS).

    Do you really want to have an xml file that contains

    10,000 file names in it?

    Since I don't anticipate having to deal with those, I'm not certain why it would be an issue for me.

    I've gotten drivers to make 10.0 and 9.1 db

    connections, but they're "unsupported". Pre-10.1

    drivers need to come stock with OEA, and be

    supported, otherwise PSC risks making OEA

    un-usable with pre-10.1 database systems.

    There were a lot of changes to the JDBC drivers

    between 9/10/10.1. In fact the sql statements to

    manipulate 4gl tables didn't really even exist in 9

    so you could only really do queries against data.

    At least that would be something though.

    In 10.1A a lot of statements were added to the

    sql and the database was modified to allow 4gl

    table modifications by sql. Older databases can't

    handle the newer statements, nor can the drivers.

    So basically db nav becomes read only on anything

    except 10.1.

    Ok. So the drivers have restrictions on pre-10.1 platforms. At least one could connect to them, nav around, and do other lookup work. Limited functionality's better than nothing at all.

    Database navigator is based on an open source product

    net.sourceforge.sqlexplorer. Its been rebranded, but

    a lot of our changes have been contributed back to

    the current maintainer. You'll notice in the JDBC

    Drivers preference page that there are place holders

    for other databases. The plugins for these other

    drivers are available ( we don't ship them for

    obvious reasons) and can be used if you have the

    drivers for them.

    Are there any docs that would help someone find these drivers?

    >> This will be a real a problem when the license

    >> policy which allows shops to run both OEA and

    >> OES expires and shops are supposed to declare

    >> which package they're going with, or paying for

    >> a new license to keep them both of them active

    >> and under maintenance. Personally I think OEA

    >> and OES should be packaged together until OES

    >> eventually dies on the vine.

    All of OpenEdge Studio with the exception of Dynamics

    and the SlickEdit control (and a few other small

    bits) are shipped with OEA 10.1a.

    So any Dynamics users are left hanging in the breeze?

    I'm sure the functionality's there, but can I just get the old OES desktop w/out all the OEA "other" stuff for when I'm doing non-project experimenting and the like?

    Tim Kuehn

  • There is a lot we want to do in this area to make it better.

    10.1B would be nice ...

    Consulting in Model-Based Development, Transformation, and Object-Oriented Best Practice  http://www.cintegrity.com

  • Well, someone though it would be a good idea.

    The "assistance" preference says that CA is floating

    text while entry, while auto completion is ".", ":".

    (There's also "ctl+space will still work").

    Eclipse default for context completion/asistance whatever you wanna call it is invoked by ctrl+space which brings up the selection list with all the nice entries in it. There is a second kind called auto completion that is unique to the oe text editor that is trigged on space, dot, colon and I think parenthesis. This one tries to guess based on unambiguous alphabetical collation of things it knows about to auto finish what you started to type (e.g. type "DEF" and you get DEFINE.

    yeah....again this is a case of eclipse provided

    functionality that we get for free.

    It needs an upgrade.

    You want an upgrade to a free feature? Sheesh.

    They're mostly skeletons for various ABL declarations

    that write the definition in a standardized format

    that I can then fill-in-the-blanks with.

    Good candidates for wizards.

    Can templates be chained? Ie, one template invokes

    another one - maybe something like:

    &DVCH\n&DVI\n

     * admin dreams of recursion.

    to declare a character var and then an integer var?

    Sure. Type "DVCH" followed by the enter key. Then type "DVI" followed by the enter key :). As many as you want.

    I'll log a bug to have this added.

    Importing a directory tree into a project is a copy

    operation - to where?

    From the source into the project directory. Right click on a file and choose "properties" in OEA. It'll tell you where it physically exists in the file system.

    Two types of imports. Import existing project just creates a pointer that project in the workspace...doesn't copy. Import files copies the files from their existing location into the selected destination folder in a project.

    Eclipse functionality...not ours.

    Or

    1. file->new->file/folder

    2. select advanced button

    3. fill in file/folder location

    4. will then show up in your project as a linked resource.

    Draw backs to this method is that OEA 10.1a doesn't handle them particularly gracefully. 10.1B is better at it. Most SCM tools choke and die with linked resources. e.g. clearcase goes hates them so much it disables all clearcase functionality for the entire project until you remove the linked resource. CVS is ok with them though.

    Are there any docs that would help someone find these

    drivers?

    http://sourceforge.net/projects/eclipsesql would be a good place to start.

    So any Dynamics users are left hanging in the

    breeze?

    Dynamics was put into the OEA in 10.1B beta if I recall correclty.

    I'm sure the functionality's there, but can I just

    get the old OES desktop w/out all the OEA "other"

    stuff for when I'm doing non-project experimenting

    and the like?

    Any OpenEdge installation (on windows at least) allows you to customize what you install. The default option is to install just about everything as far as I know.

    The windows shortcuts that get installed in the start menu, include all the old links such as oe client, oe studio, openedge desktop, as well as OEA.

    Tim Kuehn

    If this is Eclipse's doing, it's - well - lame and

    annoying. It should expand the Editor branch if it's

    not there already, put the cursor there, and

    highlight it. Not filter everything else out.

  • If this is Eclipse's doing, it's - well - lame and

    annoying. It should expand the Editor branch if

    it's

    not there already, put the cursor there, and

    highlight it. Not filter everything else out.

    Well, someone though it would be a good idea.

    Someone always does...

    >> The "assistance" preference says that CA is

    >> floating text while entry, while auto completion is ".",

    >> ":". (There's also "ctl+space will still work").

    Eclipse default for context completion/asistance

    whatever you wanna call it is invoked by ctrl+space

    which brings up the selection list with all the nice

    entries in it. There is a second kind called auto

    completion that is unique to the oe text editor that

    is trigged on space, dot, colon and I think

    parenthesis. This one tries to guess based on

    unambiguous alphabetical collation of things it knows

    about to auto finish what you started to type (e.g.

    type "DEF" and you get DEFINE.

    Which is what I'm already used to. It's just not working with the OEA I've got.

    yeah....again this is a case of eclipse provided

    functionality that we get for free.

    It needs an upgrade.

    You want an upgrade to a free feature? Sheesh.

    Ahem - I - and a lot of other ABL people - pay for maintenance, which includes some part of your salary.

    >> They're mostly skeletons for various ABL

    >> declarations that write the definition in a

    >> standardized format that I can then fill-in-the-blanks

    >> with.

    Good candidates for wizards.

    I don't care for or need "wizards", particularly if the blanks can be filled with names used previously in the code. Also, I tend to take existing templates and extend them to be something else, or chaining them together.

    Can templates be chained? Ie, one template invokes

    another one - maybe something like:

    &DVCH\n&DVI\n

     * timk519 dreams of recursion.

    >> to declare a character var and then an integer

    >> var?

    Sure. Type "DVCH" followed by the enter key. Then

    type "DVI" followed by the enter key :). As many as

    you want.

    I trust you're just being cute.

    I'll log a bug to have this added.

    Thanks.

    Importing a directory tree into a project is a

    copy

    operation - to where?

    From the source into the project directory. Right

    click on a file and choose "properties" in OEA.

    It'll tell you where it physically exists in the

    file system.

    So the file I 'imported' and then wanted to delete wasn't really where I thought it was? I'll have to double check that.

    >> I'm sure the functionality's there, but can I

    >> just get the old OES desktop w/out all the

    >> OEA "other" stuff for when I'm doing non-

    >> project experimenting and the like?

    Any OpenEdge installation (on windows at least)

    allows you to customize what you install. The

    default option is to install just about everything as

    far as I know.

    The windows shortcuts that get installed in the start

    menu, include all the old links such as oe client, oe

    studio, openedge desktop, as well as OEA.

    I'll have to check that out when I upgrade my computer.

    Thanks!

    Tim Kuehn

  • Theo:

    We looked really hard at the Visual Studio vs Eclipse question. In fact, we did a whole point-by-point comparison study on everything from features to extendability to business model.

    In the end, it came down to this: Microsoft was going to lock us in both from a money perspective and a platform perspective. In hindsight (keep in mind that we made the decision two years ago now), I still think we made the right decision. Eclipse continues to get wider adoption and richer funcationality, and the plug-market is starting to heat up. We're working on the UI thing so that people can develop MS desktop UI's without having two development environments - give us some time on that one. And yes, we do have quite a few shops that develop in java and ABL. In fact, we are getting quite a list of customers outside the U.S. that are doing pure Linux deployments, desktop and all. Our goal is to not impose a platform preference on our customers and partners, and we felt that Eclipse was more in keeping with philosophy then Visual Studio. Yeah, it is java underneath, but don't read that as us endorsing a platform.

    Another consideration in our research: The development environment contains quite a few "hidden" costs in licensed technologies and the like. We're really trying to reduce those to give us more flexibility in how we go to market. The VS stack was going to make that problem worse, not better. When working on a .NET desktop UI, there will be additional development costs - almost no way around that (controls and the like). But we didn't want to impose that on everyone in the base.

    The final point that pushed Eclipse over the line was the fact that two of our other product units were settling on Eclipse as well. Obviously that helps us internally, but more important we want to integrate in their tools and create some development synergy between the various products. In theory, that could be done with either environment, but in practice it is a lot easier with Eclipse.

    I'll get to some of the other points made on this thread when I get more time. But thanks, everyone, for jumping in both with questions and answers. And keep it up!

  • Theo:

    We looked really hard at the Visual Studio vs Eclipse

    question. In fact, we did a whole point-by-point

    comparison study on everything from features to

    extendability to business model.

    Those are all valid arguments for the Progress company. The question remains: is it also the best suite for an ABL-developer? I would definitly have no counter arguments when ABL would be as consistent for the front end choice: meaning support for a Java front end as deep as the .Net front end. But there is no such thing (yet).

    For me it's very disappointing how the UI-designer is integrated in Eclipse: "an AppBuilder hack should be good enough for the next year(s)" seems to be the message. Sure, I like the solution as a technical guy, being able to parent MFC-views to the Eclipse canvas. But it must look very strange to new developers.

    For me the most important response to the "likes and dislikes" question therefor is: increase the release cycle frequency of the development environment and pick up the pace. Do something about the UI-designer, get rid of the AppBuilder, move forward, make choices!

    A good example of not moving forward fast enough is the SQL-engine: it's still not possible to do anything to the schema via DDL. An example from the docs: "You cannot drop the first index created on a table, except by dropping the table". Why is this? Why is the database focus always performance, performance, performance and not feature completion?

  • Eclipse default for context completion/asistance whatever you

    I do think that we have muddied this topic somewhat so that I'm not always sure that we are talking about the same thing. One type is called "code assist". From the on-line help:

    Source to be both editable and extenseable. Those I think would be more appropriate for large blocks of code such as Tim is suggesting. Is there any reason this functionality couldn't point to a selection box which read from a list that one could add to? Not to mention needing them to be editable.

    Creating our own wizards is a further level beyond this where some interaction is required. Do you anticipate making this open and easy?

    Dynamics was put into the OEA in 10.1B beta if I recall correclty.

    Are there aspects of OES which are not fully incorporated into OEA in 10.1B?

    Consulting in Model-Based Development, Transformation, and Object-Oriented Best Practice  http://www.cintegrity.com

  • For keyboard templates, it would appear that the two complaints are my annoyance at the difficulty of switching to lower case keyword standard and Tim's of wanting recursive definitions.

    NO - I don't want recursition, I want to be able to chain their calls - one template fires off another, etc. much like function calls.

    No recursion please - that implies variables and exit conditions, which aren't generally what I think go with templates.

  • Those are all valid arguments for the Progress company. The question

    remains: is it also the best suite for an ABL-developer?

    If one starts with the assumption that no choice is going to be perfect for everyone, I think that one can make a very strong case for the Eclipse choice having been the best one because:

    1) it doesn't lock us in;

    2) there is an increasingly strong set of tools available for the platform; and

    3) it provides the hope and expection of all of the Progress family of products supported under a single IDE.

    Against this it seems the only real downside is that the .NET development environment might not be as strong as it might have been another way, but that other way would have significantly impacted the above advantages in a negative way. Presumably, you must be a .NET guy and so the mix doesn't look as attractive to you, but from a greater good perspective, I think the case is very strong.

    Consulting in Model-Based Development, Transformation, and Object-Oriented Best Practice  http://www.cintegrity.com

  • Even if it isn't full recursion, some of the same issues arise.

    Other than a small incremental cost at setup, what is the issue about having DVCH, DVIN, and DVSPECIALCOMBO (presumably something shorter like DVCB) to cover this requirement?

    For anything longer like this, I would much rather have an extensible list of templates to choose from and perhaps a single keystroke that brought up the selection list.

    Consulting in Model-Based Development, Transformation, and Object-Oriented Best Practice  http://www.cintegrity.com

  • DVCH et al are fine for small things like defining char vars and the like.

    However, a I go along, I'm running into recurring patterns of code.

    Those patterns get turned into a template.

    Then I run into repeatedly doing the same sequence of templates.

    Unfortunately, I can't easily turn that pattern into a pattern of templates, and I'm concerned that some of the templates may get too big to easily fit into an editor.

    It's also harder to maintain patterns of templates that are stored as full-length text strings. It's also harder to form new patterns of templates.

    I could get by with a way to fire off a sequence of templates rather being able to embed child template calls in parent template instances.

    I'm not sure what the issue is - all this that should need to be done would be to define a way to start a template expansion, and a recursive call to the template generator.

  • Perhaps some more concrete examples would make it clearer, but ...

    It seems to me that the keyboard template technology is fine for the little stuff, as you noted. It might be the wrong thing to try to turn that into something else.

    The Source template technology, on the other hand, I believe is fairly rich, although the examples currently in OEA are fairly simple. I believe that these are based on JET and are capable of some fairly sophisticated processing as a part of the expansion, e.g., where a single template could produce a number of different looking results depending on the values supplied at generation time. But, I haven't really looked into the details since some of them seem to be locked down and there is no obvious mechanism for expanding to new instances within OEA.

    You might look at the files I pointed to in the PEG summary I referred to earlier in the thread and see what you find. I think this might be a more suitable direction than trying to overload the keyboard templates.

    Consulting in Model-Based Development, Transformation, and Object-Oriented Best Practice  http://www.cintegrity.com

  • The biggest issues with OE for us are:

    - large business rules procedures (over 5000 lines) take way to long to save (rebuild the outline section ???) making it virtually useless since we have lots of those. (existing bug logged with PSC).

    - the buffer field auto-completion is non existant, so one of the major benefits of OE goes down the drain.

    wish list:

    - make OE a plug-in for Eclipse in a distributable form (ie, I had to hunt down the OE plug-ins in order to copy over my existing full SDK Eclipse installation)

    - extend the auto-complete to work with variable names, procedures names, (both internal and external), etc... look at how Eclipse for java does things and copy everything that is possible.

    - Make the "open procedure" the same way as Java does it. CTRL-Click. Do that for include files as well as procedures on "run" statement. both internal and external procedures (again, just copy how java works).

    - auto-format for indentation. We have lots of legacy code that doesn't comply with our current coding standards. a SHIFT-CTRL-I to reformat the identation of all the code would be great.

    Thanks

    Jonny Oenning

    Linx ADS

  • 4 things I would like to see:

    1. The ability to insert Objects, Events & Preprocessors in the code by using the right-mouse-click, just like you can in the section editor.

    2. Show variables datatypes in the outline view.

    3. Show parameters and variables that are defined in Internal Procedures in the outline view.

    4. Show defined parameters in the outline view.

  • By far the thing most needed in OEA -

    A real Web environment tool. The old Wizards from WebSpeed were horrible. Yet that's pretty much what's there today, left over from day one - we're talking 10 years. A small change to make the report use an SDO is hardly fitting a product that has garnered quite a bit of PSC revenue.

    FFW added a bunch of good functionality on the server side. But the development tool was nothing that could ever be realistically tackled - based on both the size of the effort, and the interest PSC had. Now that a world class development framework is employed, some world class web plug in's are available. Connecting those ala PHP/JSP with an AJAX toolkit to WebSpeed would finally move the product forward.